KR102084933B1 - Proximity sensor with nonlinear filter and method - Google Patents

Abstract

A sensor for a portable connection device comprising a filter 30 is arranged to reduce noise components on the sampled input signal, and the filter is arranged to take into account only input measurements that are systematically changed in the same direction, thus determining a predetermined time window The output value is updated when all of the input samples in E are greater than or less than the current output value, and the current output value is repeated when the input samples in the time window are less than or greater than the current output value.

Description

Proximity sensor and method with nonlinear filter {PROXIMITY SENSOR WITH NONLINEAR FILTER AND METHOD}

The present invention relates to a smart proximity sensor in combination with a processor for processing the output of the proximity sensor and a processor arranged to output a signal identifying between proximity to the human body and proximity to inanimate objects. The present invention particularly relates to, but is not limited to, a connected portable device having such a smart proximity sensor and arranged to adapt the RF emitted from the radio interface to maintain a specific absorption rate (SAR) within a given limit.

Capacitive proximity detectors are used in many modern portable devices, including mobile phones and tablets, to determine whether the device is close to a user's body part. This information is important in several ways: it is used to detect whether the telephone is being actively operated by the user and whether the user is looking at the display, in which case the displayed information can be adapted and / or Switch the device from the low power state to the active state. Importantly, this information is used to adapt the power level of the radio transmitter to comply with legal SAR limits. Capacitive proximity detection is also used in touch-sensitive displays and panels.

Known capacitive sensing systems measure the capacitance of an electrode and detect an increase in capacitance when the device is placed in close proximity to the human body (eg, hand, head, or knee). The variation in sensor capacity is relatively small and often corresponds to a few percent of the "background" capacity seen by the sensor when no conductive body is in proximity. Known capacitive detection systems include a digital processor for subtracting drift and noise contributions and can deliver a digital binary flag indicating proximity in real time based on the digital value of the unique user's capacity and / or a programmable threshold.

It is therefore an object of the present invention to provide a method for identifying inanimate objects in a capacitive proximity detector that overcomes the above limitations.

The invention will be better understood with the aid of the description of one embodiment given by way of example and illustrated by the drawings.

1 shows schematically a capacitive proximity sensor of a portable connection device.
2 is a flow chart illustrating a nonlinear filtering method.
3 shows a noise signal processed by a moving average filter and by a nonlinear filter.
4 shows the standard deviation in relation to the filter window size.
5 is a flow chart illustrating a selective baseline estimation method.
6 illustrates an optional multiple threshold identification method.

Although FIG. 1 schematically illustrates a capacitive proximity detector in a connected portable device such as a portable telephone, laptop computer, or tablet, it should be understood that the filters and methods of the present invention may be applied to a variety of applications.

The detector is sensitive to the capacitance Cx of the electrode 20 which will increase slightly in the proximity of the user's hand, face or body. Fluctuations due to body proximity are blurred by the self-capacitance of the electrode 20, which in turn is not stable. The capacitive signal is preferably amplified and processed by analog processor 23, which can also subtract a programmable offset, and converted to raw digital values by A / D converter 25. Sample R (n) may be encoded as a 16 bit integer, or in any other suitable format.

The raw sample R (n) also contains noise and unwanted disturbances attenuated by the filter 30 in a non-ideal world. The filter 30, which will be described in more detail below, provides a series of samples U (n) useful for processing in successive stages.

Unit 60 is a baseline estimator that generates a series of samples A (n) that approximates the instantaneous values of the baseline taking into account drift. This is then subtracted from the U (n) sample of difference unit 40 and provides a drift correction sample D (n). The identifier unit 50 then generates a binary value 'PROX' indicating proximity of the user's hand, face, or body. However, the invention is not limited to binary outputs, but also includes a detector that generates a multi-bit proximity value.

In an optional variant of the invention, the baseline estimator 60 measures the variation of the proximity signal at a determined time interval, such that the logical PROX value is asserted and the variation is not in the predetermined acceptance region or is variable, as represented in FIG. 5. And a drift compensation unit arranged to track and subtract drift from a proximity signal that generates a drift compensation signal by freezing the tracking of the drift when in this predetermined freezing area.

An important aspect of the method of the invention is the estimation of the variation of the useful signal U _n (step 120). The variation is represented by the amount Δ _var preferably calculated in each new useful sample U _n (step 105). A possible way of estimating the variation in U _n is the difference between the sample and the previous sample, Δ_var = U _n -U _n-1 or preferably in a suitable window (U _n -U _n-1 ), for example U is the moving average of the last eight received samples of _n . However, Δ _var may represent any other suitable estimator of variation.

In step 122, the method checks whether the proximity signal is active, i.e., if there is an indication that the conductor is close. If the result of this test is positive, the method tests whether the variation Δ _var is in the predetermined receiving area (step 130). In the example, Δ _var is compared with the lower threshold t ₍₋₎ which may be negative and in most cases the upper threshold t ₍₊₎ which will be positive.

If the variation is in the receiving area, the method of the present invention treats it as a drift and updates the baseline estimate to track it (step 160). The new value for A _n can be calculated by adding the variation value Δ _var to the previous value, or in any other way.

On the other hand, if the variation Δ _var is not in the predetermined receiving region t ₍₋₎ , t ₍₊₎ , the method of the present invention regards it as the movement of the telephone and / or the user and not the drift. In step 140, the previous value A _n-1 of the baseline estimate is copied into a new one A _n . In this way, the baseline estimate A _n is frozen to a constant value.

Optionally, as shown in this example, the baseline estimate A _n may be frozen based on the variation Δ _var even when the proximity signal is not active. This is the case of the illustrated example, where the variation Δ _var in step 135 is compared with another threshold t _inact . If the variation exceeds this value, the baseline estimate is frozen (step 140), otherwise the value of A _n based on the samples U _n , U _n-1 ,... In any suitable manner. Is updated (step 148). In a possible implementation, A _n may be set equal to U _n-1 , or the average of past U _n samples.

Optional steps 170 and 180 prevent the value A _n from exceeding the value of U _n , thereby ensuring D _n > 0.

If the capacitive proximity sensor is part of a portable device connected for SAR control, the sensor electrode 20 will preferably be placed proximate to the transmit antenna of the RF transmitter to accurately determine the distance from the radio source. The sensor electrode 20 may be implemented by a conductor on a printed circuit board or a flexible circuit board, and may have guard electrodes on the back and side to suppress detection of the body and objects on the back and side of the device.

In the same application, the capacitive electrode 20 may serve as an RF antenna, or part thereof. 1 illustrates this feature of the invention. The electrode 20 is connected to the radio transmitter and receiver unit 90 via a decoupling capacitor Cd and has an inductor Ld, or another RF blocking element, for blocking the radio frequency signal. Otherwise, the radio unit 90 can in this case be connected to a separate and independent antenna from the sensing electrode 20, which can be directly connected to the analog interface 23 without the decoupling inductor Ld.

6 illustrates another optional variation of the present invention, where capacitance can be detected as a function of time as the detector approaches the inanimate (peak 650) after reaching the body (peak 630). . In the identifier 50, the capacitive signal is compared with four threshold values: t _p is the lowest and at a value indicating a sufficient distance from any body part that the transmitter can operate at full RF power when it is not exceeded Corresponds. The highest threshold t _b indicates that when it is exceeded, the antenna is very close to the body part and the power should be reduced. The intermediate thresholds t _tl and t _th define a range of bands of values that can be produced by body parts or inanimate objects, and decisions are made based on variations in the signal in these bands. According to one aspect of the present invention, when a capacitance is included between t _tl and t _th , the identifier unit checks the variation of the capacitance over time to determine whether the capacitance signal is stable. Stability can be determined, for example, by verifying that the maximum and minimum values of the signal within the determined time window are no more separated than a given value, or in any other suitable manner.

The detector of the present invention can generate the following logic signals, typically represented by PROX, BODY, and OBJECT:

-Set when PROX, C> t _p . This corresponds to the logic flag generated when the identifier unit 50 of FIG. 2 is a conventional identifier;

-Set when BODY, C> t _b ;

OBJECT, set when t _tl <C <t _th and C is stable.

The power of the RF transmitter is determined in consideration of these flags, and in particular, the flag TABLE is used as an indicator that the object whose capacity has been increased is inanimate, and that the power does not need to be reduced. In a possible implementation, if the trigger levels t _p , t _b , t _tl , t _th are in the order represented in FIG. 6, the RF power level is represented by the following table covering all possible combinations of PROX, OBJECT, and BODY. Can be given.

Preferably, filter 30 implements a nonlinear noise suppression algorithm, which will now be described with reference to FIG. Entry point 305 is executed for all raw samples generated by ADC 25. The filter unit may include hardwired logic blocks, programmed logic, or any combination thereof.

The filter 30 is configured to take into account only raw measurements R (k) proceeding systematically in the same direction, so that all input samples R (k) within a predetermined time window have a current output value U (k-1). Update output value U (k) when greater than or less than On the other hand, if the input value R (k) in the same time window is less than and larger than U (k-1), the output value does not change.

In a possible implementation represented by the flow chart of FIG. 2, filter 30 calculates and maintains two variables min (k) and max (k), which are R (k) in the window of N preceding samples. Is the minimum and maximum of (step 320).

Where N is a selectable parameter that roughly determines the width of the filtering window. N may be included between 4 and 20 in a typical implementation. For N = 8, the simulation gave satisfactory results in both noise reduction and sensitivity to small distance changes. N can be a predetermined value built into the filter, a programmable amount settable by the host system, or a dynamic value.

)에서의 최대 값보다 크다고 발견되면, U(k)의 새로운 값은 그 최소치로 설정되고(단계(362)), 각각 최대치로 설정된다(단계(364)). 단계(350 및 370) 중 어느 것도 만족되지 않으면, 값(

)은 부분적으로 U(k-1)보다 크거나 부분적으로 그 미만에 있고 출력은 이전 값으로부터 변경되지 않는다(단계(366)). 그 다음, 사이클은 R(k)의 새로운 연속적인 값이 생성될 때 반복된다(단계(305)). 정보를 가진 독자는 U의 초기 값이 이들 재귀적 단계에 의해 결정되지 않지만 예를 들면, R(0)과 같은 U(0)을 랜덤 값, 또는 단순하게 0으로 설정함으로써 필터가 초기화될 때 많은 방식으로 생성할 수 있음을 이해할 것이다.In steps 350 and 370 the values of min (k) and max (k) are compared with the final determination of the output of the filter U (k-1), where U (k-1) is less than the minimum value, or window Samples (

If found to be greater than the maximum value in &lt; RTI ID = 0.0 &gt;), then the new value of U (k) is set to its minimum value (step 362), and each is set to the maximum value (step 364). If neither of 350 and 370 is satisfied, the value (

) Is partially greater than or partially below U (k-1) and the output does not change from the previous value (step 366). The cycle then repeats when a new consecutive value of R (k) is generated (step 305). The reader with the information says that the initial value of U is not determined by these recursive steps, but many times when the filter is initialized by setting U (0), such as R (0) to a random value, or simply 0, for example. It will be understood that it can be created in a way.

3 represents a simulation of the nonlinear filter of the present invention. Line 145 is a representation of a simulated signal (rectangular pulse) with a significant amount of white noise superimposed. Line 149 represents the same signal after processing with a moving average filter with a linear filter: window length (N = 8). Line 147 finally shows the output of the filter of the present invention which also has a window length (N = 8). While both the linear filter and the linear filter of the present invention stand out of the signal from noise, it will be understood that the variability of the signal 147 is significantly lower than the variability of the linear filter 149.

This can be explained by recalling that the standard deviation of the output of the linear filter decreases only along the square root of the window length (or pass bandwidth). The filter of the present invention is arranged to strongly reduce the probability of output variation due to statistical variation. For example, suppose you take an input R consisting of a constant value due to superimposed noise, and assuming that the output U (k-1) is perfectly in the middle of the noise-free value of R at a given moment, N consecutive The continuous output value U (k) will only change if the input sample is on the same side of the center value; The probability of change is then (1/2) ^N-1 .

A characteristic of the filter of the present invention is that although small transient changes may not affect the output, the output of the linear filter will change, but will hardly change. This may be considered a disadvantage in some applications, although the output of a linear filter composed of small signals overwhelmed by noise may not actually be useful.

The advantage of the filter of the present invention is that it provides a strong reduction of noise fluctuations in a simple algorithm. When the filter is applied to the proximity detector of FIG. 1, it is relatively simple to find a threshold for the identifier 50 that provides a reliable proximity trigger with a minimal false signal. This will be more difficult through a simple average. In order to achieve the same result, the linear average filter must increase N and increase the expansion of the window. However, this measurement will increase the computational burden, reduce the bandwidth and thus reduce the sensitivity to fast transients.

In the frame of the proximity detector of FIG. 1, the nonlinear filtering processor of the present invention may completely replace the low pass filter block 30 or may be used in combination with a linear digital filter before and after. As represented or also after that, it may be inserted before the baseline correction units 40, 60.

법칙을 따르는 평균 필터(410)에서보다 빠르게 감소한다는 것을 이해할 것이다.4 shows noise reduction versus number of samples, assuming the input R consists of a constant value with superimposed noise. Noise is expressed as the standard deviation of the output signal. In the nonlinear filter 430 of the present invention,

It will be appreciated that the decrease is faster than in average filter 410 following the law.

The filter of the present invention is capable of several improvements and adaptations. For example, in one variation, U (n) can be changed to max (k), but not exactly, and min (k), respectively, but can be shifted towards the limit by a very small fraction predefined. For example, the update at steps 362 and 362 can be replaced by:

Where a represents a predefined coefficient between 0 and 1.

According to another variant, the filter compares the average change in the input signal with the expected spread of the measurement, and updates the output value only when the average change is greater than the predetermined portion of the expected spread, and thus is important.

20: electrode
23: analog processor
25: A / D Converter
30: filter
40: tea
50: identifier
60: baseline estimator
90: receiver
105: new sample (U (n))
120: estimation of variation
122: test for proximity
130: test variant for lower and upper thresholds
135: Test for inactivity threshold
140: freeze deformation
145: simulated signal
147: filter output
149: output of linear filter
160: update baseline estimate
170: Test A (n) against U (n)
180: caps A (n)
305: new R (x) value
320: Compute and maintain maximum and minimum values
350: test filter output for min (k)
362: set to minimum
364: set to maximum
366: Copies old values
370: test filter output for max (k)
410: standard deviation of the mean
430: standard deviation of the filter of the present invention
630: approach of the body
650: Inanimate approach

Claims (9)

A sensor for a portable connection device,
The sensor includes a nonlinear filter arranged to reduce noise components on the sampled input signal,
The nonlinear filter updates the output value when all of the input samples in the predetermined time window are greater or less than the current output value, taking into account only that the input measurement proceeds systematically in the same direction, and the input in the time window. And arranged to repeat the current output value when the samples are partially less than and greater than the current output value. The method of claim 1,
The nonlinear filter sets the output value to the minimum value of the input samples in the time window when the current output value is less than the minimum value of the input samples in the time window, and the current output value is the maximum value of the input samples in the time window. And set the output value to a maximum value of the input samples in the time window when greater than the value. The method according to claim 1 or 2,
The nonlinear filter moves the output value to the minimum value of the input samples in the time window when the current output value is less than the minimum value of the input samples in the time window, and the current output value is determined by the input samples in the time window. And move the output value to a maximum value of input samples in the time window when greater than the maximum value. The method according to claim 1 or 2,
A sensor for a portable connection device comprising a linear low pass filter. The method according to claim 1 or 2,
A sensor for a portable connection device, comprising a baseline correction unit and an identifier. The method according to claim 1 or 2,
The sensor is a capacitive proximity sensor arranged to determine whether a user is in proximity to the portable connection device based on the capacitance indicated by the sensing electrode. In a portable connection device,
A sensor according to claim 1 or 2 arranged to determine whether the user is in proximity to the portable connection device based on the capacitance indicated by the sensing electrode,
And the portable connected device is operably arranged to adapt to the RF power of a radio transmitter based on the user's proximity to the portable connected device. The method of claim 7, wherein
And the sensing electrode is also an emitter of a radio wave. delete

Images